Solve and Netmaster are using two different logs:
1. Sysout (Hardcopy) log produced from the region joblog through LOG1/LOG2/.../LOGx
2. Online or activity log produced from the VSAM datasets through NMLOG1/NMLOG2/.../NMLOGx
In some cases, there are several minutes delay between when SOLVE is started and
the messages found from the activity log, causing some messages lost and eventually impacting monitoring procs running at that time.
Why do initial messages not found from the activity log since the region start?
Release: SLACCS00200-5-SOLVE:Access-Session Management
At the time of the region start, the activity log is implemented through following messages:
* Action $NM LOGFILES parameter group
-> IAIN0126 Processing for parameter group $NM LOGFILES started
* Allocate and Open Activity Logs (Online)
-> ALLOCATE DD=NMLOG01 DSN=xxxxx.xxxxxx.NMLOG01 DISP=SHR
UDBCTL OPEN=NMLOG01 ID=NMLOG01 LSR SIS DEFER
ALLOCATE DD=NMLOG02 DSN=xxxxx.xxxxxx.NMLOG02 DISP=SHR
UDBCTL OPEN=NMLOG02 ID=NMLOG02 LSR SIS DEFER
* Start LOGPROC
-> Sysparm LOGPROC=$LOPROC
* Notification of Activity log start
===> LOG ENTRY: LOPROC01 ACTIVITY LOGGING COMMENCING
The Sysout log receive all messages as its sysout is already allocated at the region startup.
The Activity log only receive messages after VSAM datasets allocation and OML procedure start.
A delay is introduced before getting the activity log active. It is caused by the calculation of the statistics performed by the logproc to produce the display for /LOGDAYS (Date List for Domain xxxx).
The delay depends on the number of records in all Activity logs and may be variable.
The problem can be circumvented by the following setting :
- Use or Create $NMPARMS member in TESTEXEC library
This member is called before NMINIT in the startup if it exists and ensure
that any command or parameters in this member are the first things that
the region will do when it starts - so it is better than adding the line
- Insert $GLBL $LOGDAYS NO or &&000$LOGDAYS = NO
- On next region restart, the /LOGDAYS shortcut will display 0 records for
all lines, indicating the calculation of number of log days is dropped.
This should speed up the start of logging and reduce the delay observed